100 - NASA€¦ · launched through the bolt or stud being measured. These errors are caused by...

14
United States Patent Gieman et al. [19] illIlllll III I|11 Illl iH lm lib illBIlIHIII IIll I| US005970798A 7_J___ (7/_,_- _ I111 Patent Number: '[ 5,970, 98 t45] Date of Patent: Oct. 26, 1999 [54] ULTRASONIC BOLT GAGE [75] Inventors: Stuart M. Gieman, Titusville; Geoffrey K. Rowe, Winter Park, both of Fla. [73] Assignee: The United States of America as represented by the Administrator of the National Aeronautics and Space Administration. Washington, D.C. [21] Appl. No.: 08/936,788 [22] Filed: Sep. 25, 1997 [51] Int. Ci. 6 ...................................................... F16B 31/02 [52] U.S. C! ................................................................. 73/761 [58] Field of Search ........................................ 73/761. 649 [56] References Cited U.S. PATENT DOCUMENTS 3,969,810 7/1976 4.062,227 12/1977 4,308,751 1/1982 4,363242 12/1982 4.471,657 9/1984 4,640,131 2/1987 4,818,470 4/1989 5,029,480 7/1991 5,156.050 10/1992 5,161,594 11/1992 5,220,839 6/1993 5A04.643 4/1995 5,440,929 8/1995 Pagano ...................................... 29/407 Heyman .................................... 73/630 Thumer et al ............................ 73/627 Heyman .................................... 73/761 Voris et al ............................ 73/761 X Kromng et al ........................... 73/600 Richardson et al ..................... 376'245 Kibblewhi_ .............................. 73/761 Schmid et al ......................... 73/761 X Bolton et al .......................... 73/761 X Kibblewhite .............................. 73/761 Rice .......................................... 73/761 Huang et al .............................. 73/628 5A99.540 3/1996 5_571,971 11/1996 5,584,627 12/1996 5,663,503 9/1997 5,675,087 10/1997 5,717,143 2/1998 5,807,048 9/1998 Whaley et al ............................ 73/761 Chastel et al ............................. 73/761 Ceney et al ........................... 73/761 X Dam et al ................................. 73/649 MacLauchlan et al ................... 73/761 Jenco et al ................................ 73/761 D'agraives et al ................... 73/761 X OTHER PUBLICATIONS Leisk. G.G., et al., "Digital Computer Algorithms to Calcu- late Ultrasonic Wave Speed", Materials Evaluation. 840-843. (July 1996). Prima D. Examiner--George Dombroske Assistant Examiner--Jewel V. Thompson Attome); Agent, or Firm---Beth A. Vrioni; Gary G. Borda [57] ABSTRACT An ultrasonic bolt gage is described which uses a cross- correlation algorithm to determine a tension applied to a fastener, such as a bolt. The cross-correlation analysis is preferably performed using a processor operating on a series of captured ultrasonic echo waveforms. The ultrasonic bolt gage is further described as using the captured ultrasonic echo waveforms to perform additional modes of analysis. such as feature recognition. Multiple tension data outputs. therefore, can be obtained from a single data acquisition for increased measurement reliability. In addition, one embodi- ment of the gage has been described as multi-channel, having a multiplexer for performing a tension analysis on one of a plurality of bolts. 18 Claims, 7 Drawing Sheets 110 i ////// ¢ 100 PULSER 112 RECEIVER 113 https://ntrs.nasa.gov/search.jsp?R=20000030657 2020-04-26T07:13:09+00:00Z

Transcript of 100 - NASA€¦ · launched through the bolt or stud being measured. These errors are caused by...

Page 1: 100 - NASA€¦ · launched through the bolt or stud being measured. These errors are caused by apparent changes in amplitude, peak splits, peak merges, and distortion in return waveforms

United States Patent

Gieman et al.

[19]

illIlllllIIII|11IllliHlmlibillBIlIHIIIIIIllI|US005970798A 7_J___ (7/_,_- _

I111 Patent Number: ' [ 5,970, 98

t45] Date of Patent: Oct. 26, 1999

[54] ULTRASONIC BOLT GAGE

[75] Inventors: Stuart M. Gieman, Titusville; GeoffreyK. Rowe, Winter Park, both of Fla.

[73] Assignee: The United States of America as

represented by the Administrator of

the National Aeronautics and Space

Administration. Washington, D.C.

[21] Appl. No.: 08/936,788

[22] Filed: Sep. 25, 1997

[51] Int. Ci. 6 ...................................................... F16B 31/02

[52] U.S. C! ................................................................. 73/761

[58] Field of Search ........................................ 73/761. 649

[56] References Cited

U.S. PATENT DOCUMENTS

3,969,810 7/1976

4.062,227 12/1977

4,308,751 1/19824,363242 12/1982

4.471,657 9/1984

4,640,131 2/1987

4,818,470 4/1989

5,029,480 7/1991

5,156.050 10/1992

5,161,594 11/1992

5,220,839 6/1993

5A04.643 4/1995

5,440,929 8/1995

Pagano ...................................... 29/407

Heyman .................................... 73/630Thumer et al ............................ 73/627

Heyman .................................... 73/761Voris et al ............................ 73/761 X

Kromng et al ........................... 73/600Richardson et al ..................... 376'245

Kibblewhi_ .............................. 73/761Schmid et al ......................... 73/761 X

Bolton et al .......................... 73/761 X

Kibblewhite .............................. 73/761

Rice .......................................... 73/761

Huang et al .............................. 73/628

5A99.540 3/1996

5_571,971 11/1996

5,584,627 12/1996

5,663,503 9/1997

5,675,087 10/1997

5,717,143 2/1998

5,807,048 9/1998

Whaley et al ............................ 73/761Chastel et al ............................. 73/761

Ceney et al ........................... 73/761 XDam et al ................................. 73/649

MacLauchlan et al ................... 73/761

Jenco et al ................................ 73/761

D'agraives et al ................... 73/761 X

OTHER PUBLICATIONS

Leisk. G.G., et al., "Digital Computer Algorithms to Calcu-

late Ultrasonic Wave Speed", Materials Evaluation.

840-843. (July 1996).

Prima D. Examiner--George Dombroske

Assistant Examiner--Jewel V. Thompson

Attome); Agent, or Firm---Beth A. Vrioni; Gary G. Borda

[57] ABSTRACT

An ultrasonic bolt gage is described which uses a cross-

correlation algorithm to determine a tension applied to a

fastener, such as a bolt. The cross-correlation analysis is

preferably performed using a processor operating on a series

of captured ultrasonic echo waveforms. The ultrasonic bolt

gage is further described as using the captured ultrasonic

echo waveforms to perform additional modes of analysis.

such as feature recognition. Multiple tension data outputs.

therefore, can be obtained from a single data acquisition for

increased measurement reliability. In addition, one embodi-

ment of the gage has been described as multi-channel,

having a multiplexer for performing a tension analysis on

one of a plurality of bolts.

18 Claims, 7 Drawing Sheets

110i

//////¢

100

PULSER112

RECEIVER113

https://ntrs.nasa.gov/search.jsp?R=20000030657 2020-04-26T07:13:09+00:00Z

Page 2: 100 - NASA€¦ · launched through the bolt or stud being measured. These errors are caused by apparent changes in amplitude, peak splits, peak merges, and distortion in return waveforms

U.S. Patent Oct. 26, 1999 Sheet 1 of 7 5,970,798

110100

PULSER112

1RECEIVER _ 113

/ FIG. 1A

SIGNAL

0114J _ _ _ I_ 116

% f _ fY V

A B

TIME

FIG. 1B

Page 3: 100 - NASA€¦ · launched through the bolt or stud being measured. These errors are caused by apparent changes in amplitude, peak splits, peak merges, and distortion in return waveforms

U.S. Patent oct. 26, 1999 Sheet 2 of 7 5,970,798

1 SIGNAL

AA B

Lr

T

r TIME

FIG. 2A

T

A

t, /

Iooo

A

FIG. 2B

I

I

I

,,/// ,^^

I VvII

Vf

v

B

F

FIG. 2C

i T

_r

Page 4: 100 - NASA€¦ · launched through the bolt or stud being measured. These errors are caused by apparent changes in amplitude, peak splits, peak merges, and distortion in return waveforms

U.S. Patent Oct. 26, 1999 Sheet 3 of 7 5,970,798

AIIII

T1

E1

/

I ,^AAA^,I'vVVV' ,W1

v

TIME

(NANOSECONDS)

TO

!

!

T2v

E2

/_^AAA^,I

I 'vVVV',W2

TIME

(NANOSECONDS)

FIG. 2D

ITO

I

IAT

I

III

I.^AAA^,,'vVVV' ,W1

2E

I

IIII

] .^AA^A,,'_VV_',W2

Page 5: 100 - NASA€¦ · launched through the bolt or stud being measured. These errors are caused by apparent changes in amplitude, peak splits, peak merges, and distortion in return waveforms

U.S. Patent oct. 26,1999 Sheet4 of 7 5,970,798

120_ ULTRA. 1TRANS

1 t128 t MULTIPLEX II_

t

TEMP. t130TRANS

PULSER122

124- _ WAVE

- [ CAPTURE

FIG. 3

126

PROCESSOR

120-..,-

144

CONTROLLEDMUX

PDX-934

142c,

OSC.

FIG;. 4

146

148t

RS-422

152

GPIB IEXT

TGPIB

EXT

I

COMPUTER

Page 6: 100 - NASA€¦ · launched through the bolt or stud being measured. These errors are caused by apparent changes in amplitude, peak splits, peak merges, and distortion in return waveforms

U.S. Patent Oct. 26, 1999 Sheet 5 of 7 5,970,798

0

I

, ,} I-

_' /o, , , , , , , , , , , , , , , , , , , / °

0 0 00000 0000000000 000 0 O00 0 00000 0000000000 000 O0o o ddodd 6ddddddddd ddd dogO C.O',¢'0,,I 0 gO _ _t" 0,10 gOUld"d" _10 GO_.0 '_'C',I C',,,I"_"I") ro I'01'_ I'0 C',,lC,,I C,,I C,,I 0,1,-_-,,- ,,--- I I

($8-I)NOISN3.L

tO

da

IJ_

Page 7: 100 - NASA€¦ · launched through the bolt or stud being measured. These errors are caused by apparent changes in amplitude, peak splits, peak merges, and distortion in return waveforms

U.S. Patent oct. 26, 1999 Sheet 6 of 7 5,970,798

.212fq_r_ J

208.._ _ - 210

224-

--..,

222 224

2oo FIG. 6B

Page 8: 100 - NASA€¦ · launched through the bolt or stud being measured. These errors are caused by apparent changes in amplitude, peak splits, peak merges, and distortion in return waveforms

U.S. Patent Oct. 26, 1999 Sheet 7 of 7 5,970,798

300

302 312

308 310

306

--304

FIG. 7

Page 9: 100 - NASA€¦ · launched through the bolt or stud being measured. These errors are caused by apparent changes in amplitude, peak splits, peak merges, and distortion in return waveforms

5.970.798

1

ULTRASONIC BOLT GAGE

ORIGIN OF THE INVENTION

The invention described herein was made in the perfor-

mance of work under a NASA contract and is subject to the

provisions of Section 305 of the National Aeronautics and

Space Act of 1958, Public Law 85-568 (72 Stat. 435:42

U.S.C. 2457)

TECHNICAL FIELD OF THE INVENTION

The present invention relates generally to bolt gages and

in particular the present invention relates to a system and

method of measuring the tension placed upon a bolt usingultrasonics,

BACKGROUND OF THE INVENTION

Reliable and accurate bolt tension gages are an essential

tool in the Kennedy Space Center (KSC) operational envi-

ronment to determine the amount of preload in critical bolts

and studs that are located in both ground support equipment

and flight hardware. Experience at KSC has shown that

commercially available ultrasonic bolt gages have problems

taking required measurements, Existing bolt gages perform

adequately in most instances, but can produce unacceptable

errors and uncertainties when performing tension measure-

ments in some configurations of flight hardware. Theseerrors and uncertainties often result from interference due to

multiple return waveforms from an ultrasonic pulse

launched through the bolt or stud being measured. Theseerrors are caused by apparent changes in amplitude, peak

splits, peak merges, and distortion in return waveforms

being analyzed and the bolt gage's inability to compensate

for these phenomena. Other factors that affect the accuracyof a measurement include the tensioning process which may

cause changes in geometry and material characteristics,

temperature changes between unloaded and loadedconditions, contamination between the transducer and the

surface of the item under test, contact pressure between the

transducer and the item under test, and changing or remov-

ing and replacing a transducer during the process.

Most commercial bolt gages operate by determining the

time-of-flight (TOF) of an ultrasonic return pulse in a bolt or

stud. The TOF is measured by selecting a single feature of

the item under test. Typically, an ultrasonic transducer 110 ismounted at one end of the item under test, such as a bolt 1_.

a pulse is launched through the item by a pulser circuit 112

simultaneously, the electronic clock is started. The time is

measured for a return signal, or echo pulse as detected by a

receiver 113 (See FIG. 1A). The traditional gage uses a

single waveform feature such as the slope and trigger level

setting as the requirement for stopping the clock. This single

feature TOF is the basis for comparison and measurement.The difference between an unloaded (zero applied tension)

and loaded TOF measurement for a particular holt or stud is

compared in terms of nanoseconds. Since each tension state

has a unique sound speed associated with it, the nanosecondsinterval can be converted to an ultrasonic length. Provided

there is no permanent deformation in the item under test, the

preload of the bolt or stud may be calculated either by usingHooke's Law or using a pre-determined plot of ultrasonic

length (or TOF) versus load, taking into account the speed

of sound change in the material under load and compensat-

ing for the other factors, such as temperature.

That is, preload can be determined by correlating the firstreturn waveform of a bolt in both an unloaded and a loaded

2state. The amount of the time shift measured between these

states is proportional to the preload tension in a joint coupled

by the bolt. The two time measurements are compared todetermine how far the waveform has been displaced in time.

5 The time difference is proportional to the stretch in the bolt.This method and traditional bolt gages rely on an arbitrary

zero or starting point. The resulting TOF includes the lengthfor an equipment cable, transducer and couplant delays, as

well as any delays in the measuring device. These are allsystem related external errors and can be subtracted out.

l0 Thus. an unloaded length or TOF is obtained and recorded.In a second method, the first and second return (echo)

waveforms are correlated to produce the actual TOF in the

item under test rather than including the elements of the totalmeasurement system. The first return waveform is used as

15 the reference for the second return waveform. By consider-

ing the time between the first and second returns, theexternal variables in the first TOF method and typical bolt

gages are eliminated. A reference table for each bolt, whichplots tension versus delay time. is consulted to determine

20 bolt tension. This plot is typically slightly parabolic in shape.A problem with bolts in critical applications is that they

are usually very lightweight in comparison to the load theybear. This creates a maximum preload tension value. That is,if the bolts are over-tensioned before load. they can yield or

25 break upon application of the load. Likewise, there is also a

minimum preload to prevent a joint from separating. It iswell known that simple torque preload techniques result ina +/30% variation in tension preload, and that this is usually

unacceptable in critical joints.

30 In the 30 years since the ultrasonic bolt gage was firstinvented, many methods have been tried to increase the levelof repeatability and confidence in the accuracy of readingstaken with these instruments. When a modern ulrasonic bolt

gage and its operator are performing properly, these instru-ments can measure to an accuracy of +/- 2%. Unfortunately,

35on the all too often occasions when a gage reading "jumps

peak," a user's confidence in the data collected is signifi-cantly reduced, A significant amount of peak jump error is

a result of the reliance on one single feature to meet the samemeasurement and trigger criteria in the waveform in both

4o unloaded to loaded conditions. Thus, current bolt gages

while accurate are not reliable. For example, when a current

bolt gage is used in the critical situation such as testing boltson a space station or rocket, a reliability error of 10 to 30

percent is unacceptable. This unreliability is more than an45 annoyance, it can create an unacceptably dangerous situa-

tion. If a technician has 10 critical bolts which must be

tensioned within a particular window, and the bolt gage used

as a plus or minus 30% error one time in 10_ the techniciancannot insure that all 10 bolts are correctly tensioned.

5o See FIG. 1B where the first ultrasonic wave. A. of a slack

bolt has a first zero-crossing 114 after the signal crosses the

threshold level. Waveform B is an echo pulse of the boltunder tension. Because the waveform is deformed, in this

illustration, simply attenuated, the zero-crossing 116 follow-

55 ing the threshold level does not correspond to that of the firstwaveform, This creates substantial errors when calculating

tension depending upon the single waveform feature. This is

a common effect of tensioning a bolt.

For the reasons stated above, and for other reasons stated

6o below which will become apparent to those skilled in the art

upon reading and understanding the present specification,there is a need in the art for a more reliable ultrasonic bolt

gage.

65 SUMMARY OF THE INVENTION

The above mentioned problems with bolt gages and other

problems are addressed by the present invention and which

Page 10: 100 - NASA€¦ · launched through the bolt or stud being measured. These errors are caused by apparent changes in amplitude, peak splits, peak merges, and distortion in return waveforms

5,970.798

3

will be understood by reading and studying the following

specification. The above mentioned problems with boRgages and other problems are addressed by the present

invention and which will be understood by reading and

studying the following specification. A bolt gage is

described which uses cross-correlation of multiple ultra-

sonic echo signals to determine tension applied to a bolt. The

bolt gage can include additional analysis techniques such as

feature recognition and/or other mathematical techniques toprovide multiple modes of analysis using the same acquired

data in order to increase measurement reliability. Further, thepresent invention can be multi-channel and include a plu-

rarity of remotely spaced transducers.

In particular, the present invention describes an ultrasonic

boR gage comprising an ultrasonic transducer for coupling

to a bolt. transmitting ultrasonic signals through the bolt, and

providing echo waveform output signals. The gage includes

a processor coupled to the ultrasonic transducer for control-

ling the transmission of the series of ultrasonic signals and

receiving the output signals. The processor provides analysis

outputs indicating a tension applied to the bolt using both

cross-correlation and feature recognition analysis.

In another embodiment a multi-channel ultrasonic bolt

gage comprises a plurality of ultrasonic transducers for

coupling to a plurality of bolts, transmitting ultrasonic

signals through the bolts, and providing echo waveform

output signals. A plurality of temperature sensors are located

in proximity to and in thermal contact with the plurality of

bolts, and a processor is coupled to receive the ultrasonic

echo waveform output signals as well as the outputs from the

plurality of temperature sensors. The processor provides

analysis outputs indicating a tension applied to one of the

plurality of bolts using both cross-correlation and feature

recognition analysis. The bolt gage further includes a wave-form capturing device for storing a digital representation of

the echo waveform output signals from the plurality of

ultrasonic transducers and transmitting the echo waveform

output signals to the processor, and a multiplex circuit for

coupling one of the plurality of ultrasonic transducers to the

waveform capturing device and coupling one of the plurality

of temperature sensors to the processor.

In another embodiment, a method is described for deter-

mining a tension applied to a bolt. The method comprises the

steps of transmitting an ultrasonic signal through the bolt.

capturing a series of echo waveforms of the pulse train

transmitted through the bolt, and cross-correlating two cap-tured echo waveforms to determine the tension applied to

the bolt. The step of cross-correlating can comprise the steps

of communicating the captured series of echo waveforms to

a processor, instructing the processor to perform a series of

integrations of a product of two consecutive echo wave-

forms over a range of time shifts to identify a time-of-flight

of the ultrasonic signals, and determining a tension applied

to the bolt based upon the identified time-of-flight.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1A is a prior art ultrasonic transducer coupled to a

bolt;

FIG. 1B illustrates ultrasonic echo waveforms from the

transducer of FIG. IA;

FIG. 2A illustrates a series of echo waveforms (ultrasonic

pulse train);

FIG. 2B is an expanded view of the echo waveforms ofFIG. 2A;

FIG. 2C is a cross-correlated function of the echo wave-

forms of FIG. 2B;

4FIG. 2D illustrates a 'Tlrst echo first echo" cross-

correlation technique;FIG. 2E illustrates a "first-second echo" cross-correlation

technique;

5 FIG. 3 is a block diagram of a bolt gage according to one

embodiment of the present invention;

FIG. 4 is a block diagram of another bolt gage according

to one embodiment of the present invention;

10 FIG. 5 is a graph comparing multiple modes of data

analysis for different applied tensions:

FIG. 6A is a three dimensional view of an ultrasonic

transducer holder;

FIG. 6B illustrates one embodiment for mounting a tem-

15 perature sensor; and

FIG. 7 is a block diagram of a computer screen included

in one embodiment of the present invention.

DETAILED DESCRIFrlON OF THE

20 INVENTION

In the following detailed description of the preferred

embodiments, reference is made to the accompanying draw-

ings which form a part hereof, and in which is shown by way25 of illustration specific preferred embodiments in which the

inventions may be practiced. These embodiments aredescribed in sufficient detail to enable those skilled in the art

to practice the invention, and it is to be understood that other

embodiments may be utilized and that logical, mechanical

30 and electrical changes may be made without departing from

the spirit and scope of the present inventions. The following

detailed description is. therefore, not to he taken in a limiting

sense, and the scope of the present inventions is defined only

by the appended claims.

35 The present invention uses an entire ultrasonic waveform

to find the time-of-flight of an ultrasound echo via cross-

correlation. This approach, therefore, is not reliant upon just

one feature of the waveform, such as a zero-crossing. The

bolt gage described herein uses digital signal processing ofthe entire return waveforms, which contain more usable

information than the single feature approach presently being

used. Further. multiple software gages can be implemented

and operated from common data in the present invention to

add a redundant feature for Lmproved measurement reliabil-

ity.

Analysis Algorithms

The following is a description of several example algo-

rithms which can be implemented for calculation of ultra-50sonic time-of-flight (TOF) in a fastener. For a given fastener.

the TOF is a unique and monotonic function of the loadtension in the fastener. Thus. if the TOF is measured, the

corresponding bolt tension can be determined.

55 A recent article by the G.G. Leisk and A. Saigal, "Digital

Computer Algorithms to Calculate Ultrasonic Wave Speed",

Materials Evaluation. July 1996. described and compared

five algorithms for calculating ultrasonic wave speeds for

pulse-echo non-destructive characterization of materials.

60 The article, incorporated herein by reference, deals with

Overlap, Hilbert transform, phase-slope, cross correlation

and Cepstrum algorithms. The article compares experimen-

tal results for self consistency based on the authors' par-

ticular interpretations and implementations of the algo-

65 rithms.

The present invention is not limited to any one algorithm,

or to any particular combination of algorithms. The present

4o

45

Page 11: 100 - NASA€¦ · launched through the bolt or stud being measured. These errors are caused by apparent changes in amplitude, peak splits, peak merges, and distortion in return waveforms

5.970,798

5

invention utilizes, as much as possible, algorithms which use

the entire ultrasonic waveform as opposed to particular

features, such as peaks or zero-crossings. Other waveform

algorithms or even peaks and zero-crossings may be used asa redundant TOF measurement in order to increase the

reliability of the measurements. It is taught herein that a TOF

determined by several independent techniques, which yieldthe same results, is more reliable than any single technique'sresult.

Cross Correlation Analysis

In the present invention, a "cross-correlation" of two

signals A(t) and B(t) can be easily and simply expressed as:

F(r)=f_+_A(t_B(t_r)dt I1)

where A(t) is a first echo waveform and B(t-x) is a second

consecutive echo pulse offset by time "c. Because F(x) is

maximum when z=T (see FIG. 2C). the TOF can be deter-

mined by varying z until F('0 is maximized. Contrary to

Equation (1), the function is not integrated over infinity, just

over the time windows of interest. For example, a windowcan be used from x=-0 to T, where 10,1" Iincludes the first and

second echoes in a pulse train. To illustrate this method,

consider how a computer will calculate the function F(x). At

time "c--O, if the timing window is chosen properly, F(0)

should be very close to zero, since A(t) and B(t) have no

overlap and they individually integrate to zero. As the timingsteps of x increase. B is slid backwards in time until it starts

to overlap A. When the first peak of B coincides with a last

peak of A, the correlation integral will have its first positive

peak, Later in x, the last peak of A will match the first valley

of B, the last valley of A will match the first peak of B. and

the correlation will have its first (negative) valley. The

process continues until B has slid in "tuntil it matches A as

exactly as it can. The match may not be exact, due todistortion, but the correlation will have its global maximum

peak. As x continues to increase, the correlation will con-

tinue to oscillate, going through successively smallermaxima until it goes back to flat line zero. The x for which

the correlation function is maximum is the TOE

Two specific methods for performing cross-correlationare described herein with reference to FIGS. 2D and 2E. The

first method is referred to as 'Tlrst-echo first-echo" correla-tion. In this method a waveform for the bolt in its initial

tension state (usually slack, i.e. zero tension) is captured and

stored. The portion of the waveform of interest is the firstecho bounce. A window Wl is defined around the echo

waveforrm and a time T1 is measured from when the initial

pulse was transmitted (T=0) and the start of the window, asillustrated in FIG. 2D. The waveform is assumed to be zero

everywhere outside of the window, even though in reality

the waveform is not zero. For analysis purposes, the wave-form E1 in the window Wl and time T1 are stored for latex

use. At sometime later, the tension of the bolt may have

changed. The same process, assuming the same transducer.cable, ect.. is used to capture both another waveform E2 in

a second time window W2 and its associated time T2 for thetensioned bolt. In many instances, time T1 and time T2 are

equal or can be manipulated to be equal (eliminate timecomponents contributed due to cables, etc.). The correlation

algorithm is used to determine how far signal E2 must bemoved earlier in time to get a maximum correlation betweenthe two waveforms, E1 and E2. This time shift is used to

identify the tension on the bolt.

6

The advantage of this technique over feature recognitionis that it works even if the waveform E2 is so distorted that

feature recognition methods would peak jump, and produce

large errors. This technique uses all the information in the

5 waveform, not just a particular feature which may or maynot be available when the tension is changed. As indicatedherein, because the waveform information is available,

additional techniques such as feature recognition can be

performed in conjunction with the cross correlation tech-l0 nique.

The second cross correlation technique is referred toherein as "first-second echo" correlation, In this technique,illustrated in FIG. 2F, two windows are defined in the same

pulse train. That is, a first window is defined around a first15 echo El, and a second window is defined around a second

echo E2. These echoes can be consecutive or located any-

where in the pulse train relative to each other. Thus, theechoes do not have to be the first and second echoes of the

pulse train. The time T required to shift signal E2 to2o maximize correlation between the two waveforms is deter-

mined. This time, therefore, is used to identify a tension in

the bolt. The advantage of this cross correlation method is

that variables in the test setup such as communication cablesare eliminated.

25 Ultrasonic Bolt Gage System

Referring to FIG. 3. a block diagram of one embodiment

of an ultrasonic bolt gage system is described in detail. The

ultrasonic system includes an ultrasonic transducer 120. a

30 device for generating ultrasonic pulses 122, and a waveform

capturing device 124. These waveforms are communicated

over a general-purpose interface bus to a processor 126. or

computer, which can be remotely located. An optional

multiplex circuit 128 is used to provide pulse signals to, and

35 received waveforms from a plurality of bolts to form a

multi-channel system. A temperature transducer 130 is pro-

vided in close thermal proximity to each bolt 132 under test.

The output from the temperature transducers can also be

multiplexed to the processor.

40 In operation, the pulse generator is coupled to an ultra-

sonic transducer attached and sonically coupled to a bolt tobe tested. An ultrasonic signal is transmitted through the

tested bolt. Return echo waveforms from the ultrasonic

signal are captured and transmitted to the processor. The

45 temperature of the bolt, or joint coupled by the bolt, is also

transmitted to the processor. The waveforms and tempera-

ture data are processed on the processor to determine thetension on the bolt. Multiple techniques, or modes, for data

analysis are provided. These techniques include, but are not50 limited to, cross-correlation and at least one feature recog-

nition. The cross-correlation technique provides an alterna-

tive method of measuring the amount of stretch as a result

of the tensioning process. To determine a tension on a bolt.

the processor is instructed to perform a cross-correlation, for

55 example by series of integrations of a product of twotime-windowed echo waveforms over a range of time shifts

to identify a time-of-flight of the ultrasonic signals. The

tension applied to the bolt is determined based upon theidentified time-of-flight. Further, at least one feature recog-

60 nition analysis can use either consecutive echo pulses in

combination with a look-up table, or a stored un-tensionedecho waveform can be used in the analysis. The feature

recognition can be zero-crossings or other waveform fea-tures. Further, other techniques such as the "Hilbert

65 transform", "Cepstrum'or other mathematical techniquescan be used as an additional method of identifying a time-

of-flight.

Page 12: 100 - NASA€¦ · launched through the bolt or stud being measured. These errors are caused by apparent changes in amplitude, peak splits, peak merges, and distortion in return waveforms

5.970.798

7

It will be understood that the processor provides analysisoutputs indicating a tension applied to a bolt using bothcross-correlation and feature recognition analysis in

response to a series of instructions executed by the proces-

sor. The processor can operate in response to firmware, orsoftware to determine the tension on a bolt as derived from

the time-of-flight of an ultrasonic pulse where the time-of-flight is determined by a multiplicity of different techniquesincluding cross-correlation in conjunction with (checked by)

feature recognition, rectification, and smoothing, or the like.It will be appreciated by those skilled in the art that the

present invention can be equally applied to numerous fas-teners referred generally herein as bolts. These fasteners caninclude, but are not limited to, bolt-nut combinations, singleor double ended bolts, and threaded bolts or machine screws

used without a nut. Additionally. this instrumentation can beused to determine tension in other applications such as trussrods in various structures, mine ceiling rods. and otherstructural applications.

EXAMPLE

In one embodiment illustrated in FIG. 4. a commercially

available ultrasonic bolt gage, such as a Raymond Engineer-

ing PDX934 bolt gage 140 is used for providing pulse

signals to the ultrasonic transducer 120. To capture the

waveforms from the ultrasonic transducer, a digital oscillo-

scope 142 is used, Further. a microprocessor controlled

multiplexer subsystem 144 and two serial (RS-422) links

1,16 to a remote computer are provided for multiplexed

communication with multiple ultrasonic transducers. In one

embodiment, a Dallas 5000 TM mioroprocessor is used with

other micro-circuitry as the microprocessor controlled mul-

tiplexer subsystem. Extra multiplexing connections 143 are

illustrated for connecting to multiple ultrasonic and tem-

perature transducers. The computer, in one embodiment, is

a Macintosh computer operating LabView software fromNational Instruments.

The gage was desired to be controlled from 400 feet away

from the space station "node" test article for safety reasons

during pressure testing. The system also needed to remotely

switch the gage from bolt to bolt, and to read the tension inthe bolts as accurately as possible. The PDX934 was cali-

brated and was used to make test measurements as part of

the improved system, but a vastly improved system wasdeveloped which incorporated our cross-correlation method

of determining the time of flight of an ultrasonic pulse. The

bolt gage 140 was used, in one embodiment, as a pulser forthe ultrasonic transducer 120.

One of the links 146 was for the computer to allow remote

adjustment of the gage 140 (delay. gain, and threshold

settings). This allows the usual and necessary tuning of the

bolt gage to acquire reliable data. The other link passed the

output from the gage to the computer. The digital oscillo-

scope 142 was added to the system at the location of the gage

(i.e., near the test article) to capture the ultrasonic signals

and recognition gate waveforms. These waveforms weretransmitted to the computer via an IE,EE-488 parallel link

150 (with exteaders 152). As long as the waveform was

available, the accuracy of the system could be improved by

using a cross-correlation technique and also by implement-

ing a better feature recognition bolt gage or two in software,

and then comparing the outputs of these gages. When the

gage outputs are all consistent, a good tension reading ismade. If the numbers from the different gages do not match,

a problem exists which can be resolved by voting, or

averaging, or otherwise adjusting as described below.

The example system provided a cross-correlation gage

and two different feature recognition gages which used the

8

PDX934 as an ultrasonic pulser. The present inventionimproves the ultrasonic bolt gage technology in at least twoways: 1) by using a cross-correlation technique, and 2) by

combining the oross-correlation technique with the multiple

5 feature recognition techniques. A sample of the data from the

pressure test is provided in FIG. 5.

FIG. 5 shows output from the cross-correlation gage 322

and two standard bolt gages 320, 324 plotted against load for

the carbon steel bolt. FIG, 5 also shows a comparison of a

10 classic PDX934 bolt gage 326 for the carbon steel bolt. The

cross-correlation bolt gage reads TOF in nanoseconds and

the standard gages read to one-ten thousandth of an inch.

Either unit can be converted to bolt tension in pounds. What

is illustrated is the change in tension with node pressure

15 along the X-axis. The node was pressurized from 0 to 15.2

PSI, and then back again. It was noted that the standard bolt

gage 326 output was not very stable in comparison to the

present invention. That is, as the node pressure between

plates secured by the bolt increased, the standard gage

2o provided inaccurate tension readings. In the example

illustrated, the problem was not peak jumping, but simple

waveform distortion. The cross-correlation technique pro-

vides better resolution and precision, uses the same

transducers, cabling, and fixtures as standard bolt gages and25 does not require the additional support equipment or field

adjustments required for the standard gages. Therefore, timedomain cross-correlation between the entire transmitted

waveform and the entire received waveforms can be pro-cessed digitally to determine bolt length and bolt tension.

30

Ultrasonic Transducer Mounting Fixture

Referring to FIGS. 6A and 6B, one example of an

ultrasonic transducer coupled to a bolt is provided. Two

35 flanges 200 and 202 are coupled together using a bolt 204and nut 206 configuration. The bolt illustrated is inserted

through the flanges from the bottom such that the nut

engages the bolt on the top of flange. Some of the threads of

the bolt will remain exposed above the nut (not shown) after

40 the nut is tensioned to the bolt. A transducer holder 208 isthreaded onto the bolt via the exposed threads. In a preferred

embodiment, the transducer holder has an access window

210 provided in its side wall. Extending from the top of the

transducer holder is a clamping mechanism 212 which

45 provides pressure between the ultrasonic transducer and theend of the bolt. It will also be understood, that an adhesive

or gel couplant can be used to attach the bolt in the ultrasonic

transducer to more reliably couple the ultrasonic signals tothe bolt.

5o A temperature sensor 220 is mounted in close proximity

to the bolt 204. The temperature sensor is necessary because

the speed of sound is used in data analysis and is dependent

upon the temperature of the bolt, as well as stress in the bolt.

Therefore, the temperature sensor is coupled to a flange

55 using an adhesive system and an insulator. For simplicity,

the temperature sensor can be taped 222 to the flange and

foam tape 224 used as the insulator, see FIG. 6B. It will be

appreciated, that the cables coming from both the ultrasonic

transducer and in the temperature sensor should be strain-60 relieved to reduce stress on the ultrasonic transducer and the

temperature sensor.

The above described mounting system is but one method

of coupling the ultrasonic transducer and temperature sensor

to a bolt for testing. Those skilled in the art can devise65 numerous variations to achieve the desired result of detect-

ing ultrasonic echo signals transmitted through a bolt anddetecting a temperature of the bolt during test. These numer-

Page 13: 100 - NASA€¦ · launched through the bolt or stud being measured. These errors are caused by apparent changes in amplitude, peak splits, peak merges, and distortion in return waveforms

5.970.798

9

ous variations are contemplated in the present invention, and

the above described system is not intended to be limiting.

Computer Screen Display

Referring to FIG. 7, one example of a computer display 5

screen 300 providing output for a user operating the ultra-

sonic bolt gage is illustrated. The screen includes an adjust-

ment region 302 for the pulse generator 122 so that variables

such as gain, threshold, and gate holdoff can be adjusted

remotely from the computer 126. The screen also provides 10

a visual illustration 304 of the echo waveforms captured

during test. Further. a magnified image 306 of one of theecho waveforms is also provided. Pointers 308 and 310 are

provided, and overlap the magnified image to illustrate bothnegative going zero crossing, and positive going zero cross- 15

ing. These pointers are preferably different colors such as

red and green. Again, these pointers are determined based

upon the feature recognition executed by the computer, anddo not have to be zero crossing points. In addition, the screen

provides bolt tension data 312 from multiple analysis tech- 20

niques. That is. tension data from multiple feature recogni-

tion analysis, and cross-correlation analysis are illustrated. It

is anticipated that during operation the various analysis data

will closely match. If any one of the data points variesgreatly from the other data points, an error in determining 25

the bolt tension can be avoided. For example, in a standard

bolt gage using a single feature recognition point, side

effects of the physical bolt tensioning process could createerrors which would be illustrated as an anomaly on the

computer screen, but would not be solely relied upon. Thus, 30

an error which would have been experienced using standard

bolt gages is avoided using the present invention.

Data Interpolation and Error Reduction35

Interpolation can be used to augment the lower resolution

of existing commercial digital data acquisition systems, as

the present sampling rates are at 100 MHZ and are inad-

equate to achieve the required accuracies. Test results show

that the waveforms do not contain frequency components 4ogreater than 10-15 MHZ, making it possible for interpola-

tion techniques to estimate missing data points by suitable

curve fitting. By doing this estimation, the probable errorband is increased and must be taken into account.

A voting/vector matching operation is also contemplated 45

to reduce errors. In this approach and by example, the first

three or four zero-crossings of the ultrasonic echo are timed

and compared with the initial tension state zero-crossing to

calculate the delta time for each of the zero-crossings. If the

delta times all agree, a more reliable timing result can be 50

obtained. If the delta times do not agree, several courses of

action may be taken. For example, a vote among the set ofresults can be taken and the value that least agrees can be

discarded, or the values can be averaged to produce a result

that is closer to the best estimated value than the worst case 55

of the sample values.

Conclusion

An ultrasonic bolt gage has been described which uses a

cross-correlation algorithm to determine a tension applied to 60a fastener, such as a bolt. The cross-correlation analysis is

preferably performed using a processor operating on a seriesof captured ultrasonic echo waveforms. The ultrasonic bolt

gage is further described as using the captured ultrasonicecho waveforms to perform additional modes of analysis, 65

such as feature recognition. Thus, multiple tension data

outputs can be obtained from a single data acquisition for

10

increased reliability. In addition, one embodiment of the

gage has been described as multi-channel, having a multi-

plexer for performing a tension analysis on one of a pluralityof bolts.

Although specific embodiments have been illustrated anddescribed herein, it will be appreciated by those of ordinary

skill in the art that any arrangement which is calculated to

achieve the same purpose may be substituted for the specificembodiment shown. This application is intended to cover

any adaptations or variations of the present invention.Therefore. it is manifestly intended that this invention be

limited only by the claims and the equivalents thereof.What is claimed is:

1. An ultrasonic bolt gage comprising:

an ultrasonic transducer for coupling to a bolt, transmit-

ting ultrasonic signals through the bolt, and providingecho waveform output signals: and

a processor coupled to the ultrasonic transducer for con-trolling the transmission of the series of ultrasonic

signals and receiving the output signals, the processor

providing analysis outputs indicating a tension appliedto the bolt using both cross-correlation and feature

recognition analysis.2. The ultrasonic bolt gage of claim 1 further comprising

a temperature sensor located in proximity to the bolt and

having an output coupled to the processor.3. The ultrasonic bolt gage of claim 1 further comprising

a waveform capturing device coupled to the ultrasonictransducer for storing a digital representation of the echo

waveform output signals.4. The ultrasonic bolt gage of claim 3 wherein the

waveform capturing device is an oscilloscope.5. The ultrasonic bolt gage of claim 1 further comprising

a multiplex circuit for coupling a plurality of ultrasonic

transducers to the processor.6. The ultrasonic bolt gage of claim 1 wherein the

cross-correlation and feature recognition analysis are per-

formed by the processor in response to a series of instruc-

tions executed by the processor.7. The ultrasonic bolt gage of claim 1 wherein the feature

recognition analysis recognizes at least two features of theecho waveform output signals.

8. The ultrasonic bolt gage of claim 7 wherein the at least

two features of the echo waveform output signals are a

positive-going zero crossing and a negative going zero

crossing.9. A multi-channel ultrasonic bolt gage comprising:

a plurality of ultrasonic transducers for coupling to a

plurality of bolts, transmitting ultrasonic signals

through the bolts, and providing echo waveform output

signals:

a plurality of temperature sensors located in proximity tothe plurality of bolts;

a processor coupled to receive the echo waveform output

signals and outputs from the plurality of temperaturesensors, the processor providing analysis outputs indi-

cating a tension applied to one of the plurality of bolts

using both cross-correlation and feature recognitionanalysis;

a waveform capturing device for storing a digital repre-

sentation of the echo waveform output signals from the

plurality of ultrasonic transducers and transmitting theecho waveform output signals to the processor; and

a multiplex circuit for coupling one of the plurality ofultrasonic transducers to the waveform capturing

device and coupling one of the plurality of temperature

sensors to the processor.

Page 14: 100 - NASA€¦ · launched through the bolt or stud being measured. These errors are caused by apparent changes in amplitude, peak splits, peak merges, and distortion in return waveforms

5.970.798

11

10. The multi-channel ultrasonic bolt gage of claim 9wherein the plurality of ultrasonic transducers are coupled toa plurality of bolts using a transducer holder threaded ontothe bolts.

11. The multi-channel ultrasonic bolt gage of claim 9 5

wherein the analysis outputs indicating a tension applied toone of the plurality of bolts is visually displayed on a

computer display screen.

12. A method of determining a tension applied to a bolt.

the method comprising the steps of:

transmitting an ultrasonic signal through the bolt;

capturing a series of echo waveforms transmitted throughthe bolt; and

cross-correlating two captured echo waveforms to deter- 15mine the tension applied to the bolt.

13. The method of claim 12 further comprising the step of:

performing a feature recognition analysis on the series of

echo waveforms to provide a second determination of

the tension applied to the bolt. 20

14. The method of claim 12 wherein the step of trans-

mitting the series of ultrasonic signals comprises the stepsof:

communicably coupling a processor and an ultrasonictransducer which is attached to the bolt; and

12

activating the ultrasonic transducer using the processor.

15. The method of claim 12 wherein the step of capturing

a series of echo waveforms comprises the steps of:

communicably coupling a waveform capturing device toan ultrasonic transducer which is attached to the bolt;and

storing the series of echo waveforms in a digital format in

the waveform capturing device.

16. The method of claim 12 wherein the step of cross-i0 correlating comprises the steps of:

communicating the captured series of echo waveforms to

a processor;

instructing the processor to perform a series of integra-

tions of a product of two time-windowed echo wave-

forms over a range of time shifts to identify a time-of-flight of the ultrasonic signals; and

determining a tension applied to the bolt based upon the

identified time-of-flight.

17. The method of claim 12 further comprising the step of

determining a temperature of the bolt.

18. The method of claim 17 further comprising the step of

communicating the temperature of the bolt to the processor.